than mercury, so, while present in trace quantities,
mercury.  From these maps, it can be seen that, for essen-
tially any given source location within the modeling
their concentrations cannot be strongly influenced by
domain, emissions of Hg(II) would result in the highest
their interactions with mercury.
deposition in the lake, followed by emissions of Hg(p).  In
are predicted to result in
Wet and dry deposition processes themselves are not
contrast, emissions of Hg0
significantly less relative deposition.
fundamentally affected by the presence of mercury and
are generally considered to be first order with respect
These results are consistent with the general understanding
to mercury concentrations.  In this context, first order
of the relative atmospheric behavior and fate of these
means that the process rate is estimated with an
different forms of mercury.  Hg(II) is very water soluble,
expression of the form rate = k • c, where k is a
with relatively strong surface adhesion properties, and is
parameter which may depend on a number of factors,
therefore much more likely to be subject to wet and dry
but which does not depend on the concentration of
deposition.  For example, Hg(II) can be wet-deposited from
mercury, c.
within precipitating clouds and even from below these
clouds, due to its very high water solubility.  Hg(p) can also
In addition, the current understanding of mercury’s
be wet-deposited relatively efficiently if its host particles
atmospheric chemistry does not include any chemical
find themselves in precipitating clouds.  In contrast to
reactions or equilibrium relations that are not first
these forms, Hg0
order with respect to mercury.
is only sparingly water soluble and
relatively volatile — thus its potential for wet and dry
Finally, vapor/particle, vapor/droplet, and droplet/soot
deposition is comparatively limited.  These fundamental
equilibrium relations can all be expressed as a ratio of
considerations are the basis for the relatively long atmo-
the concentrations in different phases.  A reasonable
spheric lifetimes estimated for Hg0 (~ 0.5 - 1 year), but
assumption can then be made that every mercury-
relatively short atmospheric lifetimes for Hg(II) and Hg(p)
(Schroeder and Munthe, 1998).  Indeed, it is often the slow
containing compound has the same proportional
atmospheric conversion of Hg0 to Hg(II) and/or Hg(p)
chance (governed by this equilibrium ratio) of being in
any given phase.  Thus, the presence of mercury from
which is required as a step to the eventual deposition of
another source is not expected to significantly affect
the interphase distribution of mercury from any other
As mentioned above, any given emissions source will
generally emit a mixture of these different forms.  For
In consideration of all of the above factors, the assumption
example, while there are significant variations based on the
that emissions of mercury from one source be considered
type of coal being burned, the type of pollution control
as independent of emissions of mercury from other sources
equipment present, and other factors, on average, coal-
appears justified.  This assumed independence is likely to
fired power plants emit a mixture comprised of approxi-
be valid for the modeling of many other trace pollutants in
mately 50 percent Hg0 , 45 percent Hg(II), and five percent
the atmosphere, but is certainly not valid, for example, for
Hg(p).  Figure 6 shows the model-estimated transfer
emissions of volatile organic compounds and nitrogen
coefficients for this average coal-combustion emissions
mixture for each of the Great Lakes.  The patterns are
similar for each of the Great Lakes, and show that, as would
Results of the Interpolation Procedures: Transfer
be expected, the propensity for atmospheric deposition
contributions is reduced substantially as the distance from
the lake increases.  Since the prevailing winds are from the
Figure 5 shows the model-estimated transfer coefficients to
west, potential contributions from regions west of a given
Lake Superior for emissions of Hg0 , Hg(II) and Hg(p).
lake tend to be greater than potential contributions from
These transfer coefficient maps show the ratio between the
similar distances east of the lake.  As another way of
understanding this phenomenon, the contribution
deposition flux to Lake Superior (micrograms of mercury
(of all forms) deposited per year per km  of lake surface)
potential falls off more steeply east of the lakes because
and hypothetical continuous emissions (grams of mercury
winds that could transport mercury from sources in these
emitted per year from a given location) of the given form of
regions occur less frequently.
mercury over the entire year 1996 from any given location
For the results presented here, a total of 84 such standard
throughout the modeling domain.  It is important to stress
source locations, many of which were clustered around the
that these maps do not incorporate emissions data.  They
basin, were used (the small circles shown in Figures 5 and
simply represent the relative propensity of emissions from
any given location for deposition in Lake Superior, based
6).  Analyses using more than 84 standard source locations
on the simulated atmospheric transport and fate of
were also performed, but the results did not vary signifi-